Re: Ziggurat code

[siko1056]

Allin Cottrell wrote
> Maybe I can give a little something back. There appears to a slight 
> mix-up in the use of *X86_32 macros in the source file in question, 
> namely randmtzig.cc. Near the top of this file there appears the 
> following preprocessor stanza:
> 
> #if ! defined (USE_X86_32)
> #  if defined (i386) || defined (HAVE_X86_32)
> #    define USE_X86_32 1
> #  else
> #    define USE_X86_32 0
> #  endif
> #endif
> 
> Here the symbol HAVE_X86_32 is coming from "elsewhere" (I presume, 
> config.h or the compiler command-line), and the symbol USE_X86_32 is 
> set conditional on that plus the compiler symbol i386. One would 
> suppose, therefore, that the code below should branch conditional on 
> USE_X86_32, but that's not the case: USE_X86_32 is not referenced at 
> all; everything below depends on HAVE_X86_32.
> 
> It seems that either (a) the code below should branch on USE_X86_32 
> rather than HAVE_X86_32, or (b) if HAVE_X86_32 is reckoned to carry 
> all the relevant information, the aforementioned preprocessor code 
> should be removed, for clarity.

Dear Allin Cottrell,

Thank you for your comment and effort! Regarding the preprocessor stanza, I
vote for option (b) and opened a bug report to remember to review this:
https://savannah.gnu.org/bugs/index.php?49305 

Best,
Kai



--
View this message in context: 
http://octave.1599824.n4.nabble.com/Ziggurat-code-tp4680058p4680065.html
Sent from the Octave - Maintainers mailing list archive at Nabble.com.